By OpenAISourceGitHub

Manuscripts from openai/math

Fork
Clone
git clone https://openversions.com/git/index/openai-math.git
Cite
Version link
https://openversions.com/index/openai-math/commit/dfd8c46870b48511e697cbb43a3c8c91d375fa2f
Version identifier (commit hash)
dfd8c46870b48511e697cbb43a3c8c91d375fa2f

Fork to create your own copy with credit to the source. Clone to get a Git copy on your computer.

All papers

Claims

Findings

An unconditional first moment for cubic Gauss sums

Unconditionally, as X→∞X \to \infty, the first moment of normalized cubic Gauss sums over all primary Eisenstein primes π≡1(mod3)\pi \equiv 1 \pmod{3} with Nπ≤XN\pi \le X satisfies ∑Nπ≤X, π≡1(mod3) primeG(π)=65c∗X5/6log⁡X+o(X5/6log⁡X)\sum_{N\pi \le X, \, \pi \equiv 1 \pmod{3} \text{ prime}} G(\pi) = \frac{6}{5} c_* \frac{X^{5/6}}{\log X} + o\left(\frac{X^{5/6}}{\log X}\right), where c∗=(2π)2/3/(3Γ(2/3))c_* = (2\pi)^{2/3}/(3\Gamma(2/3)).

For rational primes p≡1(mod3)p \equiv 1 \pmod{3}, Theorem 1.1 implies ∑p≤X, p≡1(mod3)Sp2p∼35c∗X5/6log⁡X=(2π)2/35Γ(2/3)X5/6log⁡X\sum_{p \le X, \, p \equiv 1 \pmod{3}} \frac{S_p}{2\sqrt{p}} \sim \frac{3}{5} c_* \frac{X^{5/6}}{\log X} = \frac{(2\pi)^{2/3}}{5\Gamma(2/3)} \frac{X^{5/6}}{\log X}, where Sp=∑x mod pe(x3/p)S_p = \sum_{x \bmod p} e(x^3/p) is Kummer's sum.

For every fixed integer ℓ\ell, ∑Nπ≤X, π≡1(mod3) primeθℓ(π)(G(π)−c∗N(π)−1/6)=oℓ(X5/6/log⁡X)\sum_{N\pi \le X, \, \pi \equiv 1 \pmod{3} \text{ prime}} \theta_\ell(\pi)(G(\pi) - c_* N(\pi)^{-1/6}) = o_\ell(X^{5/6}/\log X), where θℓ(z)=(z/∣z∣)ℓ\theta_\ell(z) = (z/|z|)^\ell; in particular, for every fixed nonzero integer ℓ\ell, ∑Nπ≤X, π≡1(mod3) primeθℓ(π)G(π)=oℓ(X5/6/log⁡X)\sum_{N\pi \le X, \, \pi \equiv 1 \pmod{3} \text{ prime}} \theta_\ell(\pi) G(\pi) = o_\ell(X^{5/6}/\log X).

For every fixed integer kk with 3∤k3 \nmid k, ∑Nπ≤X, π≡1(mod3) primeG(π)k=1k∈{−1,1}65c∗X5/6log⁡X+ok(X5/6log⁡X)\sum_{N\pi \le X, \, \pi \equiv 1 \pmod{3} \text{ prime}} G(\pi)^k = 1_{k \in \{-1,1\}} \frac{6}{5} c_* \frac{X^{5/6}}{\log X} + o_k\left(\frac{X^{5/6}}{\log X}\right).

At the two exceptional conductor-length configurations (P,Q)≈(Y,1)(P, Q) \approx (Y, 1) and (P,Q)≈(Y1/3,Y1/3)(P, Q) \approx (Y^{1/3}, Y^{1/3}), the second moment of A(p,q)=∑(b,e)=1β(b)N(b)iuχb(vpq2)A(p,q) = \sum_{(b,e)=1} \beta(b) N(b)^{iu} \chi_b(vpq^2) for an admissible prime convolution β\beta of length YY satisfies ∑p,q∣A(p,q)∣2≪Y7/3−η\sum_{p,q} |A(p,q)|^2 \ll Y^{7/3 - \eta} for some fixed η>0\eta > 0, uniformly for N(v),N(e)≤YδN(v), N(e) \le Y^\delta and 1+∣u∣≤Y0.361 + |u| \le Y^{0.36}.

For either ordinary (T≪X.01T \ll X^{.01}) or upper (X.01≪T≪X1/6+ρX^{.01} \ll T \ll X^{1/6+\rho}) height blocks, the prime sum with envelope V(N(n)/X)V(N(n)/X) has an exact decomposition into LO(1)L^{O(1)} Type I and bilinear terms with independent, divisor-bounded coefficients, valid simultaneously for both G(n)N(n)itG(n)N(n)^{it} and c∗μ2(n)N(n)−1/6+itc_* \mu^2(n)N(n)^{-1/6+it}, preserving prime-convolution structure for B>X.40B > X^{.40} and roughness/(SW) on ordinary blocks.

Methods

Inversion of cube completion in the level Voronoi formula via Möbius inversion, comparing the residue directly with the elementary model and bounding the tail by counting and Heath-Brown's cubic large sieve.

Source excerpts

“For squarefree r, invert cube completion by the exact identity \sum_u G(ru)W_r(Nu/U) = \sum_{(c,r)=1} \mu(c)|c|\mathcal{C}_r(U/(Nc)^3; W_r).” · Show in paper

Limiting-exponent contradiction for character moments at exceptional conductor-length configurations using a short convolution identity Λ=(2m−m∗m∗1)∗log⁡N\Lambda = (2m - m * m * 1) * \log N, functional equation shortening, and an off-diagonal Gram-matrix estimate.

Source excerpts

“Convolving with \Lambda and using 1 * \Lambda = \log N gives, on the dyad, \Lambda = (2m - m * m * 1) * \log N.” · Show in paper

Corrected dispersion with squarefree majorization and cube-frequency Poisson summation subtraction, matched against Type I comparison in the mixed term.

Source excerpts

“Opening the second moment isolates its diagonal; Poisson summation expresses the off-diagonal part in frequencies. For (5.3), the cube contributions identified below produce the model square, while the noncube contribution is small. The Type I comparison supplies the matching mixed term, giving cancellation in the corrected square.” · Show in paper

Exact bin-based stopping rule on surrogate prime products with a binomial weight (k++k−k+)−1\binom{k_+ + k_-}{k_+}^{-1} at the final bin to separate bilinear coefficients without coupled least-prime cutoffs.

Source excerpts

“For a fixed original squarefree m, exactly \binom{k_+ + k_-}{k_+} selections have the stated counts, and either all or none satisfy the crossing test. Give each selection the scalar \binom{k_+ + k_-}{k_+}^{-1}. Their sum restores the original coefficient exactly.” · Show in paper

Two-variable log-norm cell localization near the product boundary N(ab)≍X0N(ab) \asymp X_0 combined with integration by parts in Mellin height away from the boundary.

Source excerpts

“For smaller B, divide both log-norm ranges into cells. Near the product boundary, each cell has only boundedly many partners, so the localized diagonal gains the square root of the cell width. Away from the boundary, integration by parts in height supplies the saving.” · Show in paper